Demoing Madrona Labs SUMU (beta)
Madrona Labs has been teasing Sumu for quite awhile now, with an image and blog post about a year back, I’m so glad to finally get a chance to play this one up close and personally. While it is still in beta, this is not a review, but I wanted to discuss a few interesting features, sound design ideas, and answer a few of the questions you might have. I can’t go into too much detail, I have no clue about release dates, beta invitations, or pricing. As well I will not be touching on performance or stability as Sumu is still in development.
What is Sumu?
Sumu is an additive synthesizer which uses resynthesized partial files of sampled audio.
There’s a bit to unpack here, if you’re not familiar with additive synthesis, it is a method of tone generation that uses a stack of tuned sine waves to produce various tones.
In Sumu, the frequencies and amplitudes of these sine waves are determined by a resynthesized audio sample.
How does this differ from wavetable synthesis? Aren’t wavetables just a bunch of harmonics?
Yes, wavetables are harmonics, but they are just that, harmonics, they can not deviate from the harmonic series. You might not know this, but the word “partial” actually refers to partial harmonics, as in frequencies outside of the harmonic series. This allows for far more spectral motion.
Is this FFT?
FFT uses evenly spaced bands, it has a similar issue to wavetables where frequencies can’t really drift around, it’s not necessarily locked to the harmonic series, but FFT tends to have a bit of a washy blurry sound, as the more bands you use, the less timing you can account for, and vice versa. Look up uncertainty principal
What’s nice about the way Vutue/Sumu does things is that resynthesis gives frequencies their own “identity”. This means Each partial can glide up or down or wiggle independent of one another.
Why would you want additive synthesis? doesn’t’ sample playback or granular work just fine?
Once you break down a sound into sine waves, you have freedom to do all sorts of things with them, you can squeeze them closer together, spread them further apart, FM them against themselves (this is pretty unique to Sumu actually). This is where the actual fun side of Sumu comes in, as there are a variety of methods to play around with partials. For example, with granular you can play through an audio file, but with sumu you can play through an audio file at different rates per partial, allowing for some bizarre melting of audio.
I had a lot of fun streaming Sumu today, I tried to explore a variety of features, included resynthesized files, and different experimental techniques. I really enjoyed gliding the partials around and out of their natural ranges. It was fun messing with the Amen break as well, the resynthesis doesn’t quite sound identical of course, but it let me destroy the sample in new and creative ways. I still have a lot to test and experiment with but it’s great to see that it holds up for a livestream just fine. I hope this clears up a few questions you might have and more importantly I hope it inspires you and makes you excited for Sumu. Again, I’m not covering everything in this write up, and you can read through Randy’s blog for more detailed information.